it 's optional to use bad-artist embedding and you can try other textual inversions as well. PS : Consider using bad - artist embedding or boring_e621 as well. Starting the positive prompt with 'uploaded on e621' and negative prompt with '(worst quality, low quality:1.4)' help too. 2 also works but makes the style different ) For better results, it 's recommended to use it with () or you can use versions with baked-in VAE. ) Sampler : DPM ++ 2 M Karras or Euler a ( Clip skip 1 is recommended. Some of the example images here are upscaled. I personally use it with these settings : CFG : 6 - 8 steps : 23 - 150 Size : 512 x 704 or 578 x 768 ( then upscale it. It 's able to produce sfw/nsfw furry anthro artworks of different styles with consistant quality, while maintaining details on stuff like clothes, background, etc. This is my custom furry model mix based on yiffy - e18. Script for building a prompt from a lora's metadata tagsĭescription : Hello and welcome.LoCon/LoHA Training Script / DAdaptation Guide.Catbox Script for easier prompt sharing.Top 300 e621 artist wildcard (last updated November '22).'What does ControlNet weight and guidance mean?'.Different LORA sliders - what do they mean?.Crosskemono 2 (with added E621 Tag support).Crosskemono (CivitAI, links last updated: ).

AOM VAE (rename it the same as the AOM model you use).AbyssOrangeMix2 (for those without a Huggingface account).7th_furry tests (seem to be merges of the 7th layer models).

Yiffymix 2 (based on fluffyrock-576-704-832-lion-low-lr-e16-offset-noise-e1) (Use Clip Skip 1).

Legend of zelda monsters e621